1. Field of the Invention
The present invention relates to a method and system of numerical analysis for continuous data, more particularly to a method and system of numerical analysis for continuous data suitable to be implemented in hardware.
2. Description of the Related Art
Some of the algorithms are specifically dedicated to analyzing the numerical change of continuous data in preparation for the subsequent processing of the continuous data. For example, individual sections of the continuous data may be identified where each section is to be processed with a different algorithm. Prior art analysis methods for numerical change typically require searching the data forwardly and backwardly to proceed with a relevant decision.
For example, referring to FIG. 1, the numerical change of the continuous data is shown by the curve in FIG. 1, wherein the data section between time T1 and T4 is a target section to be further processed. In order to identify the target section, a prior art algorithm identifies a point of the data curve greater than a high threshold Hth (such as at time T2), and therefrom searches backwardly (as shown by the direction of the arrow D1) for another point of the data curve equal to a low threshold Lth (such as at time T1) and marks its coordinates. Then, upon moving forwardly, when a point of the data curve smaller than the high threshold Hth (such as at time T3) is encountered, therefrom the prior art algorithm searches forwardly (as shown by the direction of the arrow D2) for another point of the data curve equal to the low threshold Lth (such as at time T4), and marks its coordinates. Based on the above algorithm, the target section of the continuous data can be identified.
Although the aforementioned numerical analysis method can be implemented easily in software, the processing speed is slower and the processing throughput is lower; therefore, it may not be able to meet the requirements of real time system applications. If the aforementioned numerical analysis method is implemented in hardware, then the step of searching the data forwardly and backwardly would cause a larger temporary storage and read/write overheads. Such implementation not only increases hardware cost, but also decreases the processing throughput.
Therefore, a method of numerical analysis for continuous data without searching data forwardly and backwardly and hence suitable to be implemented in hardware is highly desirable.